Madison Kate AKA Madikat

New Flux version!
300 images used in this dataset. Using 'redhead' seems to help a little bit.

Original SD1.5 description:
Sample images made using RealisticVision model.

My best results were keeping CFG between 3 and 4, steps around 20, and LoRA weight around 0.65

Lower the LoRA weight if using in conjunction with other LoRA's; inpaint face afterwards ('only masked' option) with higher weight for most accurate results.